Abstract: Non levy Fabry-Perot interferometer(EFPI)optic fiber sensing gas is a new detection technology, with the advantages of low-cost, easy to reuse, high sensitivity and interference characteristics, it is widely used in various fields. Because the collected interference spectrum signals are non-band-limited signals, the low-pass filter can not be used to achieve non-band-limited signals denoising. In this paper, choosing wavelet transform(WT), spectrum signals are firstly made wavelet decompositions as layers and a reasonable threshold is selected to filter out the high frequency components in each layer signal, at last, the denoised interference spectrum signals are restored by the wavelet reconstruction. Online simulation and the results show that wavelet transform (WT) can be well used to achieve interference spectrum signal denoising, which improves the performance and stability of the entire optical fiber gas sensing system.